I ran some test with VDJ7 and VDJ8 with videoFPS set to 25 for both. The ideal time between calls to draw in this case is 40 milliseconds.
With VDJ7, it was pretty much right on. 40-41 milliseconds between calls.
With VDJ8, it was anywhere from 27 to 50 milliseconds between calls. It actually hit 40 milliseconds every now and then :)
So then I set the videoFPS to 60 for VDJ8. This is a fast machine I am testing on. The ideal time between calls here would be 16-17 milliseconds.
WIth that set and not full screen, I got called about every 33-36 milliseconds to draw. This is about 30 FPS though. It was consistent and smooth.
Then still with videoFPS at 60 and full screen it was bad. 1 millisecond to 60 milliseconds between calls.
Checked when drawing something and when drawing nothing. In VDJ8 with videoFPS at 30, I am seeing 17 to 50 ms but mostly at 33. It depends on fullscreen or not, skin up or minimized, etc.
With VDJ7, it was pretty much right on. 40-41 milliseconds between calls.
With VDJ8, it was anywhere from 27 to 50 milliseconds between calls. It actually hit 40 milliseconds every now and then :)
So then I set the videoFPS to 60 for VDJ8. This is a fast machine I am testing on. The ideal time between calls here would be 16-17 milliseconds.
WIth that set and not full screen, I got called about every 33-36 milliseconds to draw. This is about 30 FPS though. It was consistent and smooth.
Then still with videoFPS at 60 and full screen it was bad. 1 millisecond to 60 milliseconds between calls.
Checked when drawing something and when drawing nothing. In VDJ8 with videoFPS at 30, I am seeing 17 to 50 ms but mostly at 33. It depends on fullscreen or not, skin up or minimized, etc.
发表时间 Mon 30 Jun 14 @ 11:55 am
Note that drawing is v-synced, so on a 60Hz display, 25fps would mean sometimes 2 display refreshes and sometimes 3 are between each call.
30 fps should give quite good results though, and it looks like that is pretty much what you are seeing.
30 fps should give quite good results though, and it looks like that is pretty much what you are seeing.
发表时间 Mon 30 Jun 14 @ 12:44 pm
I figured it might have something to do with v-sync but the problem is, when the time can vary quite a bit, it can cause quite a jump in straight line movement. The videoFPS at 60 fullscreen is way off, but not good at all fullscreen. Not sure what to do about it.
At 30 FPS I see sometimes 17, sometimes 50, these can happen like every 10-15 calls. Worse when full screen.
Depending on situation you also can still get some pretty bad tearing.
Wouldn't you expect to be consistent at 60 FPS when full screen? It's way off. Near perfect when not full screen.
At 30 FPS I see sometimes 17, sometimes 50, these can happen like every 10-15 calls. Worse when full screen.
Depending on situation you also can still get some pretty bad tearing.
Wouldn't you expect to be consistent at 60 FPS when full screen? It's way off. Near perfect when not full screen.
发表时间 Mon 30 Jun 14 @ 1:04 pm
All this testing is on a fast machine.
Using the latest build 1870. When the skin is minimized, and videoFPS at 60, it is perfect smooth when not full screen. With this build and at full screen it seems more consistent but not smooth like it is when not full screen. When not full screen I am meaning I have sized the window to almost full screen.
At videoFPS of 30 it's a little better.
What the skin have to do with it? Should be nothing.
Using the latest build 1870. When the skin is minimized, and videoFPS at 60, it is perfect smooth when not full screen. With this build and at full screen it seems more consistent but not smooth like it is when not full screen. When not full screen I am meaning I have sized the window to almost full screen.
At videoFPS of 30 it's a little better.
What the skin have to do with it? Should be nothing.
发表时间 Mon 30 Jun 14 @ 2:34 pm
I rechecked everything for build 1870. It seems to be much better in this regard... Now get fairly reasonable consistent results. It's sort of hard to pin down exactly. Stand on your left foot and you get one thing, stand on right foot maybe something else. Better though.
发表时间 Mon 30 Jun 14 @ 5:36 pm
Yeah much better consistency with 1870. I see very good results with videoFPS at 60, full screen or not as long as the skin is minimized. Running video in background and my test in foreground CPU at 1 - 2 %. I added some smoothing to my code to take care of most of the spikes as well.
The main thing is the skin drawing or whatever it's doing when not minimized. At skinFPS of 30 and not minimized I get a steady jitter. So I set the skinFPS to one. When I do that, I see a jerk in the motion about once per second witch is the rate the skin is redrawing at.
You will never see what I am talking about when just displaying a video. I can pass you a test case if you want.
The main thing is the skin drawing or whatever it's doing when not minimized. At skinFPS of 30 and not minimized I get a steady jitter. So I set the skinFPS to one. When I do that, I see a jerk in the motion about once per second witch is the rate the skin is redrawing at.
You will never see what I am talking about when just displaying a video. I can pass you a test case if you want.
发表时间 Mon 30 Jun 14 @ 11:29 pm